|Home | About | Journals | Submit | Contact Us | Français|
Expression of the cancer-germline (CG) (or cancer-testis) antigen gene BORIS/CTCFL has been proposed to mediate activation of CG antigen genes in cancer. Consistent with this idea, we have observed that BORIS is frequently expressed in ovarian cancer, often in conjunction with other CG genes. Here we assessed the role of BORIS in CG antigen gene regulation and DNA methylation using normal and cancerous ovarian cell lines, and the CG genes MAGE-A1, NY-ESO-1, and XAGE-1 as models. Adenoviral vectored BORIS was expressed at robust levels and exhibited predominant nuclear localization in ovarian cells. However, BORIS expression in immortalized ovarian surface epithelial cells or ovarian cancer cell lines did not induce CG antigen gene expression or lead to CG antigen promoter DNA hypomethylation. BORIS overexpression also did not alter global DNA methylation, as assessed by genomic 5-methyl-deoxycytidine levels and LINE-1 methylation. We used decitabine to further assess the role of BORIS in CG gene activation and found that decitabine treatment induced BORIS and other CG genes with similar kinetics, suggesting that BORIS induction does not account for the induction of other CG genes by decitabine in ovarian cancer cells. In agreement, siRNA knockdown of BORIS did not block decitabine-mediated induction of CG genes or DNA hypomethylation in ovarian cancer cells treated with this agent. We conclude that BORIS is insufficient for CG antigen gene expression and DNA hypomethylation in ovarian cell lines, and that additional factors are likely required for CG antigen expression in ovarian cancer.
Cancer-germline (CG) (also known as cancer-testis or CT) antigen genes are infrequently expressed in normal adult somatic tissues but are expressed in germ cells, trophoblast, and cancer (1, 2). The inherent immunogenicity and tumor specificity of these genes have made them attractive cancer vaccine targets (1). When expressed in germ cells, CG antigens do not elicit immune responses due to the lack of HLA expression; however, when CG antigens are expressed in tumors they can elicit spontaneous immune responses in cancer patients (1). Recent clinical studies suggest that CG antigen-targeted vaccines are safe and promote immunological responses (3-5), encouraging continued clinical investigation of this approach.
Transcriptional silencing of CG antigen genes in normal somatic tissues is attributed to promoter DNA hypermethylation, while activation of these genes in cancer closely correlates with promoter DNA hypomethylation (1, 6). In agreement, CG antigen gene expression is induced by treatment with hypomethylating agents such as decitabine (DAC) and 5-azacytidine (7-9). Other epigenetic mechanisms, including histone modifications and methylated DNA-binding proteins (MBDs) also regulate CG antigen gene expression (10-14). While it is apparent that DNA hypomethylation plays a crucial role in CG antigen gene induction, it may not be sufficient to promote CG antigen gene expression in cancer in all circumstances (15, 16). Other factors, including sequence-specific transcription factors and global epigenomic regulators, may also promote CG antigen expression (15, 17, 18).
One example of a global epigenomic regulator of CG antigen genes is a recently-identified male germline-specific protein, Brother of the Regulator of Imprinted Sites (BORIS), which is expressed in several types of human cancers including epithelial ovarian cancer (EOC) (1, 17, 19-22). The expression pattern and chromosomal localization of BORIS indicates that it is an autosomal CG antigen gene (20, 23). BORIS, which is also known as CTCF-like protein or CTCFL, is a paralog of CTCF (CCCTC-binding factor), a transcription factor that plays a key role in chromatin insulation and genomic imprinting (20, 23). In addition, BORIS was recently shown to play a role in murine spermatogenesis (24). BORIS and CTCF differ in their N- and C-termini, but share a central zinc finger-binding domain capable of binding a similar spectrum of DNA sequences (20). Interestingly, BORIS and CTCF may have reciprocal expression patterns, as spermatocytes positive for BORIS are negative for CTCF (20, 23).
BORIS expression has been implicated in the activation of CG antigen genes (19, 21). In lung cancer cells, occupancy of the CG antigen gene NY-ESO-1 promoter by CTCF was associated with gene silencing, while BORIS occupancy coincided with gene activation (19). In primary human fibroblasts, exogenously expressed BORIS led to de-repression of previously silent CG antigen genes including MAGE-A1 (21). BORIS overexpression was also found to alter the DNA methylation state of specific regions of the MAGE-A1 promoter (21). However, in later studies BORIS overexpression in melanoma cell lines did not activate MAGE-A1 expression or induce promoter hypomethylation (25). In addition, melanoma tissue samples were found to express MAGE-A1 but not BORIS, calling into question the general role of BORIS in MAGE-A1 gene expression (25). More recently, glioma stem cells were observed to frequently express CG antigen genes in the absence of BORIS expression (26). Other data have questioned the frequency with which BORIS is expressed in breast cancer, presenting data discordant with an earlier report (21, 27). Based on these conflicting findings, it is clear that additional studies are required to clarify the role of BORIS in promoting CG antigen gene expression.
Recently, we have reported that BORIS is frequently expressed in ovarian cancer and have observed that the expression of BORIS coincides with expression of other CG genes in this malignancy [(22) and Woloszynska-Read et al., manuscript in preparation]. Similar coordinate expression of BORIS and other CG antigen genes has been observed in head and neck, lung, and prostate cancer (28, 29). Taken together, these observations spurred us to examine the functional role of BORIS in the activation of CG antigen genes, using cultured ovarian cells. We also sought to determine whether BORIS overexpression altered DNA methylation, either specifically at CG antigen gene promoters or globally. Finally, we investigated whether BORIS plays a functional role in the activation of CG genes elicited by pharmacological inhibition of DNA methylation with decitabine.
We have previously shown that immortalized human ovarian surface epithelial cells (IOSE121) and four ovarian cancer cell lines (OVCAR3, SKOV3, A2780, and OVCAR429) do not express BORIS or other CG antigen genes at significant levels (22, 30). However, decitabine treatment induces CG antigen gene expression in all five cell lines, indicating that DNA methylation actively represses these genes (22, 30). To determine whether BORIS expression is capable of activating CG antigen genes in these cell lines, we overexpressed BORIS using an adenoviral approach. Western blot analyses revealed high levels of BORIS protein expression upon infection with recombinant BORIS virus, at both 48 and 96 hours post-infection (Figure 1A). Overexpressed BORIS ran at or near the expected molecular weight of 75.7 kDa (Figure 1A). Although BORIS expression appeared to be reduced at 96 hours (Figure 1A), we observed no obvious phenotypic differences (e.g. altered cell growth or death) in cells overexpressing BORIS (data not shown). To determine whether BORIS protein localizes to the nucleus, consistent with its role as a transcription factor, we conducted immunofluorescence (IF) analysis in OVCAR429 cells (Figure 1B). BORIS IF analysis was performed using two different clones of BORIS-specific mouse monoclonal antibodies to confirm BORIS antibody specificity (data not shown) (31). As shown in Figure 1B, BORIS showed predominantly nuclear localization, as indicated by co-localization with DAPI nuclear staining. Nuclear BORIS expression is in agreement with its putative role as a transcriptional regulator.
To determine whether BORIS expression activates CG antigen gene expression in ovarian cells, we analyzed the expression of the prototype CG genes MAGE-A1, NY-ESO-1, and XAGE-1 by RT-PCR, following BORIS overexpression. Decitabine (DAC) treatment of these cell lines served as a positive control for CG antigen gene induction. At 48 hours post-infection, BORIS expression did not induce the expression of CG antigen genes in either IOSE121 cells or ovarian cancer cell lines, while DAC treatment induced CG gene expression, as expected (Figure 2A). Identical results were seen at 96 hours post-infection (data not shown). The small amount of XAGE-1 PCR product seen in OVCAR3 cells infected with recombinant BORIS adenovirus at 48 hours was not due to BORIS overexpression, as similar levels of PCR product were observed in the LacZ control (Figure 2A and data not shown). We also examined the effect of BORIS overexpression on CG antigen gene expression at extended times post-infection of 1 and 2 weeks. At these time points, CG antigen genes remained repressed (Figure 2B). All end-point RT-PCR results for CG genes were confirmed using quantitative real-time RT-PCR, which gave identical results (data not shown).
The data presented above suggest that BORIS overexpression may not affect CG antigen promoter methylation. However, it is possible that BORIS could partially influence the DNA methylation state of CG antigen gene promoters, but not to an extent required to induce CG gene expression. To test this, we utilized quantitative sodium bisulfite pyrosequencing to determine the DNA methylation state of the MAGE-A1, NY-ESO-1, and XAGE-1 promoters in ovarian cell lines following BORIS overexpression. Notably, we observed no significant changes in the promoter methylation levels of CG antigen genes upon BORIS overexpression in both IOSE121 and ovarian cancer cell lines (Figure 3, panels A-C). In addition, at extended times of BORIS overexpression of 1 or 2 weeks, CG antigen promoter methylation remained unaltered in ovarian cancer cell lines (Figure 3D and data not shown). These data suggest that BORIS expression is insufficient for DNA hypomethylation of CG antigen gene promoters.
BORIS expression during male germ cell differentiation correlates with the resetting of DNA methylation marks, and spermatocytes positive for BORIS show reduced or absent 5-methyl-cytosine (5mC) signal (20). While BORIS expression did not affect CG antigen promoter methylation as described above, it remained plausible that BORIS expression could impact global DNA methylation, which has not been investigated previously. To test this, we used two quantitative measures of global DNA methylation: (i) a liquid chromatography-mass spectrometry (LC-MS) assay to measure total 5-methyl-2'-deoxycytidine (5mdC) in genomic DNA (32), and (ii) a pyrosequencing assay for the LINE-1 repetitive DNA element (30). Overall, we did not detect significant changes in 5mdC in ovarian cells transduced with BORIS, with the exception of a small increase in SKOV3 cells (Figure 4A). In agreement, LINE-1 did not show significant methylation changes upon BORIS overexpression in ovarian cell lines (Figure 4B). As expected, internal controls (cell lines treated with decitabine) showed both reduced 5mdC levels and LINE-1 methylation (Figure 4). Taken together, the data indicate that BORIS overexpression does not significantly alter global DNA methylation in ovarian cells.
It was previously reported that decitabine treatment induces BORIS expression in normal human dermal fibroblasts prior to the induction of MAGE-A1, leading to the suggestion that BORIS may mediate the induction of other CG antigen genes by this agent (21). To test this scenario in ovarian cells, we treated OVCAR429 cells with 1 µM decitabine and collected RNA over a time course of 8, 12, 24, and 48 hours to use for gene expression analysis of BORIS, MAGE-A1, NY-ESO-1, and XAGE-1 (after normalization to GAPDH). Of these transcripts, only MAGE-A1 was induced at 8 hours, while MAGE-A1, BORIS, and NY-ESO-1 showed similar levels of induction at 24 hours post-treatment (Figure 5A). All four genes were induced by 48 hours post-treatment (Figure 5A). These data are inconsistent with a general role for BORIS in decitabine-mediated CG antigen gene induction.
We additionally examined the kinetics of CG antigen promoter DNA hypomethylation using pyrosequencing. All four CG antigen gene promoters showed similar kinetics of DNA hypomethylation, but there were small differences in the magnitude of DNA hypomethylation at each region (Figure 5B). It is interesting to note that MAGE-A1 expression was induced prior to detection of hypomethylation of its promoter (Figure 5, panels A and B). These data suggest either that additional effects of decitabine contribute to the early activation of MAGE-A1, or alternatively that a very small number of hypomethylated alleles can significantly contribute to MAGE-A1 induction. We additionally determined the kinetics of decitabine-mediated induction and hypomethylation of BORIS and other CG genes in additional cancer cell lines and observed similar results (data not shown). Taken together, these data suggest that decitabine-mediated induction of CG antigen genes is not dependent on BORIS.
As an additional test of the role of BORIS in decitabine-mediated CG antigen gene activation, we used BORIS-specific siRNAs to knockdown BORIS expression in decitabine-treated OVCAR3 cells. OVCAR3 cells were utilized as they displayed greater siRNA transfection efficiency and more robust BORIS knockdown relative to other ovarian cell lines (data not shown). The BORIS monoclonal antibodies available to us were unable to detect endogenous BORIS protein in ovarian cell lines (data not shown), so we assessed BORIS knockdown using quantitative real-time reverse transcriptase PCR (qRT-PCR). BORIS or control siRNAs were administered 24 hours prior to decitabine treatment, which was then conducted for an additional 48 hours. Figure 6A shows that two different BORIS-targeting siRNAs significantly diminished BORIS mRNA expression post-decitabine treatment, as compared to control siRNA. In contrast, decitabine-mediated induction of MAGE-A1 and NY-ESO-1 were not significantly affected by BORIS knockdown (Figure 6, panels B and C). Moreover, BORIS knockdown did not alter the extent of decitabine-mediated CG antigen gene promoter methylation or global DNA methylation (data not shown). These data reinforce the notion that BORIS is not required for decitabine-mediated induction of CG antigen genes in ovarian cells.
While promoter DNA hypomethylation is a key mechanism leading to activation of CG antigen genes in human cancer, the mechanism underlying this process has not been resolved (1, 15, 33). Thus a report that ectopic expression of BORIS led to CG antigen gene expression in cultured normal human dermal fibroblasts, coincident with methylation changes at the MAGE-A1 promoter, was of great interest (21). These data provoked us to assess the functional role of BORIS in CG antigen gene induction using in vitro ovarian cell models. Ectopic overexpression of BORIS in ovarian cells did not induce CG antigen gene expression, nor did it affect the methylation of CG antigen gene promoters. These observations held true for both ovarian cancer cell lines as well as an immortalized surface ovarian epithelial cell line. In contrast to BORIS overexpression, decitabine treatment induced CG gene expression in ovarian cell lines, demonstrating that these cells contain the transcriptional machinery needed to drive CG antigen gene expression, if epigenetic repression is relieved. The lack of an effect of BORIS expression on CG gene promoter methylation, and its coincidence with CG gene repression, is consistent with a strict dependence of CG gene expression on promoter DNA hypomethylation (15).
Consistent with our data, previous work showed that ectopic expression of BORIS using a plasmid vector failed to induce MAGE-A1 expression in melanoma cell lines (25). In addition, it was recently demonstrated that BORIS knockout does not prevent hypomethylation of its gene target Gal3st1 in murine spermatocytes (24). However, a recent study reported that overexpression of BORIS using a plasmid vector led to upregulation of CG antigen genes and an increase in hypomethylated CG antigen promoter alleles in an immortalized oral keratinocyte cell line (28). Interestingly, oral keratinocytes responded distinctly to ectopic BORIS expression depending on the level of overexpression, with greater effects on cell proliferation at lower levels of overexpression (28). Thus it is possible that titrating the expression of BORIS in ovarian cells could result in effects distinct from that which we report. More generally, the contradictory findings on the effects of BORIS overexpression in different cell types may be explained if the response to BORIS expression is cell context dependent. For example, other transcriptional or epigenomic regulatory factors may work in concert with BORIS to drive CG gene expression and these factors may be limiting in specific cell types (18).
In addition to its potential role in CG antigen gene regulation, it was postulated that BORIS expression in cancer may disrupt methylation boundaries imposed by CTCF and that this could allow the aberrant methylation patterns to spread, leading to hypomethylation of normally methylated DNA (34). This mechanism would presumably be anticipated to affect DNA methylation globally. We addressed this question using ovarian cells by measuring the impact of BORIS overexpression on global DNA methylation parameters. BORIS expression did not impact either total genomic 5mdC levels or methylation of the common repetitive element LINE-1. Moreover, knockdown of BORIS in the context of decitabine treatment did not alter the ability of the drug to cause global DNA hypomethylation (data not shown). Therefore, our in vitro data do not support a functional connection between BORIS expression and global DNA hypomethylation. However, our findings do not exclude the possibility that in the in vivo setting BORIS expression may either be associated with and/or contribute to the global DNA hypomethylation phenotype.
An observation made in a previous study was that decitabine treatment led to the induction of BORIS at an earlier time point (8 hours) than it did MAGE-A1 (48 hours) (21). This led to the intriguing proposal that BORIS induction may functionally contribute to decitabine-mediated induction of MAGE-A1 and other CG genes (21). Our data are not in agreement with this model. In ovarian cancer cells, decitabine treatment induced BORIS expression coincident with other CG antigen genes, and the kinetics of promoter hypomethylation of BORIS and other CG antigen genes was also very similar. Moreover, siRNA knockdown of BORIS did not significantly impact decitabine-mediated CG antigen gene induction or promoter DNA hypomethylation.
The current study may not be fully explanatory for understanding the functional contribution of BORIS to CG antigen gene regulation in ovarian cancer as it utilized in vitro model systems. We have previously shown that BORIS is relatively frequently expressed in ovarian cancer, and that its expression is closely associated with epigenetic activation of the gene by promoter DNA hypomethylation (22). Our unpublished data in ovarian cancer, as well as observations from other malignancies, suggest that CG antigen genes, including BORIS, are coordinately upregulated, in conjunction with promoter DNA hypomethylation, in human cancer (1, 15, 28, 29). In the future, it will be constructive to utilize in vivo tumor model systems to continue to dissect the potential role of BORIS in CG antigen gene regulation.
We thank Dr. Nelly Auersperg and the Canadian Ovarian Tissue Bank for IOSE121 cells, Dr. Ivan Still of Roswell Park Cancer Institute for ovarian cancer cell lines, and Dr. Bert Vogelstein of Johns Hopkins University School of Medicine for the HCT116 and DKO cell lines. This work was supported by NIH RO1CA11674 (ARK), a Cancer Vaccine Collaborative Grant from the Cancer Research Institute/Ludwig Institute for Cancer Research (KO), the Ovarian Cancer Research Fund (KO and ARK), and NIH CA16056 (to Roswell Park Cancer Institute). AWR was supported in part by NIH training grant 5T32CA009072.
Ovarian cancer cell lines (SKOV3, A2780, OVCAR3, OVCAR429) were propagated as described previously (22). IOSE121 cells (SV40 immortalized normal human ovarian surface epithelium cells) were obtained from the Canadian Ovarian Tissue Bank and Dr. Nelly Auersperg (University of British Columbia) and were grown as described previously (22). The human colorectal cancer cell lines HCT116 and DNMT1-/-, DNMT3b-/- HCT116 (DKO) cells were obtained from Dr. Bert Vogelstein (Johns Hopkins School of Medicine) and have been described previously (10, 35). Ovarian cell lines were treated with decitabine (DAC, 5-aza-2'-deoxycytidine, Sigma Chemical Company, St. Louis, MO) as described in the text.
A BORIS adenoviral expression vector was generated using ViraPower™ Adenoviral Gateway® Expression Kit, according to the manufacturer's instructions (Invitrogen, Carlsbad, CA, USA). Briefly, we cloned the human BORIS open reading frame (ORF) into the entry vector pENTR/D-TOPO (Invitrogen) using the Gateway system. Subsequently, the BORIS ORF was sub-cloned into the pAd/CMV/V5-DEST expression construct. This vector was transfected into 293A cells using Lipofectamine2000 (Invitrogen) to produce recombinant adenovirus. An adenovirus expression plasmid containing the lacZ gene was used as a negative control. Adenovirus titers were determined by plaque formation assay, and recombinant viral supernatants were stored at 80˚C. Ovarian cells were infected with recombinant virus at a multiplicity of infection (MOI) of 50, and BORIS and LacZ (β-galactosidase) expression were detected by Western blot or IF.
Western blot analyses were performed as described previously (10), using 20 µg of protein per sample. Anti-BORIS antibody has been described previously (31) and was used at a 1:20000 dilution. The anti-V5 mouse monoclonal antibody 1:5000 (Invitrogen) was used to detect β-galactosidase, and anti-α-tubulin 1:2500 (Accurate Chemical and Scientific Corporation, Westbury, NY) was used as a loading control. HRP-conjugated anti-mouse or anti-rabbit antibodies (GE Healthcare, Piscataway, NJ) were used as secondary antibodies.
Cells were grown on glass coverslips and infected with BORIS adenovirus at a MOI of 50. 48 hours post-infection, coverslips were washed in PBS, fixed in 2% formaldehyde for 15 min, and washed twice with 1x PBS. The cells were then incubated with anti-BORIS mouse monoclonal antibody (31) (1:500) in 1x PBS containing 0.2% Triton X-100 (PBS/Triton) for 1 h at room temperature. Following washes in PBS/Triton, cells were incubated with TRITC-conjugated goat anti-mouse IgG (Jackson Immunoresearch Laboratories, Westgrove, PA) (1:200) for 30 min. Coverslips were dipped for 20 seconds in 100 ng/ml DAPI solution, washed in PBS, and mounted with Aquamount (Polysciences, Inc., Warrington, PA). Samples were viewed using a Zeiss Axioskop (Carl Zeiss, Thornwood, NY) equipped with a Hamamatsu digital CCD camera (Hamamatsu Corporation, Bridgewater, NJ).
Total RNA was extracted using TRIzol (Invitrogen, Carlsbad, CA). RT-PCR and qRT-PCR for BORIS (NCBI Entrez GeneID 140690), MAGE-A1 (GeneID 4100), NY-ESO-1 (GeneIDs 1485 and 246100), and XAGE-1 (GeneIDs 653219, 653220, 653048, 9503, and 653067) was accomplished as described previously (10, 22). High homology did not allow specific quantification of NY-ESO-1 and XAGE-1 isoforms. qRT-PCR was done on triplicate samples, and expression analysis of each gene was performed a minimum of two times.
Genomic DNA was isolated using the Puregene DNA isolation kit (Gentra Systems, Minneapolis, MN), and methylation of the BORIS, MAGE-A1, NY-ESO-1, and XAGE-1 promoters was analyzed by pyrosequencing as described previously (11, 22, 30). High homology did not allow specific quantification of NY-ESO-1 and XAGE-1 isoforms. To verify efficient bisulfite conversion, non-CpG cytosines served as internal controls, and unmethylated and methylated DNAs were run as additional controls. Pyrosequencing was done on duplicate samples, and pyrosequencing assays were performed a minimum of two times.
Knockdown of BORIS in OVCAR3 cells was achieved using siRNA oligos (Ambion, Austin, TX). Two different BORIS siRNAs were utilized: 5'-GGAUCAACCUACAGCUGGUtt-3' (siBORIS1), and 5'-GGCCUGAAGGAGGAGGAAAtt-3' (siBORIS2). Cells were transfected with 50 nM BORIS siRNAs or a control non-targeting siRNA (Dharmacon, Lafayette, CO) using Lipofectamine 2000 (Invitrogen).